One of the greatest challenges for the microelectronics industry in the coming years is to identify advanced dielectric materials that will replace silicon dioxide as an inter and intra metal layer dielectric. Dielectric film layers are fundamental components of integrated circuits and semiconductors. Such films provide electrical isolation between components. As device densities increase, multiple layer dielectric films are generally used to isolate device features. When forming dielectric films it is important for the film to exhibit certain properties, such as good gap fill, thermal stability and favorable electrical properties. The most widely used dielectric layer, silicon dioxide (SiO2) is formed by a variety of methods. The most commonly used methods are chemical vapor deposition (CVD) and plasma CVD.
As device densities shrink, the gaps between lines become smaller, and the demands on dielectric films become more rigorous. When the critical feature size goes to less than 0.25 microns, the dielectric constant (κ) of the dielectric material acquires increasing importance. For example, as the industry moves to smaller interconnects and device features, the actual length of the interconnecting lines increases and the space between them decreases. These trends increase the RC delay of the circuit.
There are generally two ways to reduce the RC delay for a given geometry: (1) you can reduce the resistance of the interconnect lines by using different metals; or (2) you can reduce the dielectric constant by modifying or using different dielectric material.
Increased RC delay has a detrimental effect on the speed of the device, which has tremendous commercial implications. Further, narrower line spacing results in reduced efficiency due to the impact of higher capacitive losses and greater crosstalk. This reduced efficiency makes the device less attractive for certain applications such as battery powered computers, mobile phones, and other devices. Reducing the dielectric constant would have a favorable impact on capacitive loss and crosstalk. Thus, it is highly desirable to reduce the RC delay of the device.
Currently, devices may incorporate five or six dielectric layers, all comprised of silicon dioxide. Silicon dioxide (SiO2) has a relatively high dielectric constant at about 4.0. Replacing SiO2 with a suitable low dielectric constant (low κ) material will lead to a dramatic improvement in speed and reduction in the power consumption of the device. Such advanced low-dielectric materials would play an important role in enabling the semiconductor industry to develop the next generation of devices.
A variety of materials have been investigated as low κ dielectric layers in the fabrication of semiconductors. Fluorine has been added to SiO2 films in an attempt to lower the dielectric constant of the film. Stable fluorine doped SiO2 formed by plasma CVD typically has a dielectric constant of 3.5 to 3.7; however, significantly lower κ values are desired.
Another plasma CVD approach to create low κ films is the deposition of highly crosslinked fluorocarbon films, commonly referred to as fluorinated amorphous carbons. The dielectric constant of the more promising versions of such films has generally been reported as between 2.5 to 3.0 after the first anneal. Issues for fluorinated amorphous carbon remain, most notably with adhesion; the thermal stability, including dimensional stability; and the integration of the films.
Polymeric materials have also been investigated. For example, spin coated polymeric materials have been employed. Despite their lower κ values, these polymers are not entirely satisfactory due to processing and material limitations. Polymers are generally thermally and dimensionally unstable at standard processing conditions of about 400 to 450° C. While these materials have been considered for embedded structures, as a rule they are not suitable for full stack gap fill or damascene structures.
Because of the disadvantages of spin-coated polymers, vapor phase polymerization has been explored as an alternative method for the preparation of low κ materials. One particular class of polymers which has been prepared through vapor phase polymerization are the polyxylylenes (also known as parylenes) such as parylene N (ppx-N), and parylene F (ppx-F). Parylenes have κ values ranging from 2.3 to 2.7 and are thus attractive as low dielectric materials for use in integrated circuits. However, the parylenes that have been prepared to date exhibit poor thermal stability as with ppx-N; expensive as with ppx-F, or have issues with mechanical stability.
To date, advanced low κ materials have not been successfully employed in the semiconductor industry. As such, there is continued interest in the identification of new materials, as well as methods for their fabrication, that have low κ values, high thermal stability, are fully manufacturable and result in reliable devices that are cost effective.